Plasmid-Resolved Metagenomics for Tracking Horizontal Transfer of Antimicrobial Resistance Determinants Across Microbial Communities

The spread of antimicrobial resistance (AMR) throughout microbial communities is a serious concern for both human and environmental health, since plasmids act as the main carriers for the horizontal gene transfer (HGT) of antibiotic resistance genes (ARGs). Yet, conventional short-read metagenomics has in the past been unable to determine the extrachromosomal context of ARGs, thereby hiding the mechanisms involved in the spread of resistance. Plasmid-resolved metagenomics has now become a groundbreaking approach that brings together third-generation long-read sequencing (using PacBio HiFi and Oxford Nanopore R10.4), improved assembly algorithms, machine learning classifiers, and proximity ligation assays (Hi-C) in order to fully reconstruct plasmids directly from complex microbial communities. This article brings together the technological frameworks, bioinformatic advances, and ecological findings that are shaping this quickly developing area. We carry out a critical assessment of the long-read assemblers such as hifiasm-meta, metaMDBG, metaFlye, HiCanu, and myloasm, showing that they are capable of recovering complete circular plasmids in a variety of environmental samples. We look at the computational methods used for identifying and separating plasmids, pointing out geNomad's deep neural network design and PlasMAAG's assembly-alignment graph method based on contrastive learning, both of which allow for accurate plasmid reconstruction even in complicated metagenomes. We also consider how Hi-C-based proximity ligation helps to establish associations between hosts and plasmids, finding that plasmids carrying AMR have a much wider range of phylogenetic host species than those that do not carry resistance, and thus form well-connected transmission networks across the various One Health interfaces such as wastewater treatment systems, aquaculture sediments, and microplastic plastispheres. Although there have been major progressions, difficulties still remain in resolving plasmid microdiversity, in standardising benchmarking protocols, and in experimentally verifying computational predictions. The future integration of this approach with single-cell microfluidics and stable isotope probing will enable the direct measurement of plasmid transfer kinetics, which in turn will allow for proactive monitoring and control of the spread of AMR mediated by plasmids in interconnected ecosystems.